LossType
Enums · AiDotNet.Enums
Defines different loss functions used to measure how well a model’s predictions match the actual values.
For Beginners
Loss functions are like a “scoring system” for AI models. They measure the difference between what the model predicted and what the correct answer actually is. A lower loss means the model is making better predictions. During training, the optimizer adjusts the model’s parameters to minimize this loss value, gradually improving the model’s accuracy.
Fields
| Field | Summary |
|---|---|
LossType BinaryCrossEntropy |
Binary Cross Entropy - cross entropy specialized for two-class (yes/no) problems. |
LossType CTC |
CTC Loss - Connectionist Temporal Classification for sequence-to-sequence alignment. |
LossType CategoricalCrossEntropy |
Categorical Cross Entropy - cross entropy for multi-class classification with one-hot encoded labels. |
LossType Charbonnier |
Charbonnier Loss - a differentiable variant of L1 loss, popular in image restoration. |
LossType Contrastive |
Contrastive Loss - learns embeddings by pulling similar pairs together and pushing dissimilar pairs apart. |
LossType CosineSimilarity |
Cosine Similarity Loss - measures the cosine of the angle between prediction and target vectors. |
LossType CrossEntropy |
Cross Entropy - measures the difference between two probability distributions. |
LossType Dice |
Dice Loss - measures overlap between predicted and actual segmentation masks. |
LossType ElasticNet |
Elastic Net Loss - combines L1 (Lasso) and L2 (Ridge) regularization penalties. |
LossType Exponential |
Exponential Loss - used in boosting algorithms like AdaBoost. |
LossType Focal |
Focal Loss - gives more weight to hard-to-classify examples, reducing the impact of easy ones. |
LossType Hinge |
Hinge Loss - used for “maximum-margin” classification, especially with Support Vector Machines. |
LossType Huber |
Huber Loss - combines MSE for small errors and MAE for large errors. |
LossType Jaccard |
Jaccard Loss - also known as Intersection over Union (IoU) loss for segmentation tasks. |
LossType KullbackLeiblerDivergence |
Kullback-Leibler Divergence - measures how one probability distribution diverges from another. |
LossType LogCosh |
Log-Cosh Loss - the logarithm of the hyperbolic cosine of the prediction error. |
LossType Margin |
Margin Loss - used in capsule networks with separate margins for positive and negative classes. |
LossType MeanAbsoluteError |
Mean Absolute Error - measures the average absolute difference between predictions and actual values. |
LossType MeanBiasError |
Mean Bias Error - measures the average bias (systematic over/under-prediction) of the model. |
LossType MeanSquaredError |
Mean Squared Error - measures the average squared difference between predictions and actual values. |
LossType ModifiedHuber |
Modified Huber Loss - a smooth approximation combining hinge loss and quadratic loss. |
LossType NoiseContrastiveEstimation |
Noise Contrastive Estimation Loss - approximates softmax for large vocabulary problems. |
LossType OrdinalRegression |
Ordinal Regression Loss - preserves the ordering relationship between classes. |
LossType Perceptual |
Perceptual Loss - compares feature representations from a pretrained network. |
LossType Poisson |
Poisson Loss - appropriate for count data that follows a Poisson distribution. |
LossType Quantile |
Quantile Loss - predicts specific percentiles rather than the mean. |
LossType Quantum |
Quantum Loss - a loss function inspired by quantum computing principles. |
LossType RealESRGAN |
RealESRGAN Loss - a composite loss for Real-ESRGAN super-resolution models. |
LossType RootMeanSquaredError |
Root Mean Squared Error - the square root of the average squared differences. |
LossType RotationPrediction |
Rotation Prediction Loss - a self-supervised loss that predicts image rotation angles. |
LossType ScaleInvariantDepth |
Scale Invariant Depth Loss - measures depth prediction quality regardless of absolute scale. |
LossType SparseCategoricalCrossEntropy |
Sparse Categorical Cross Entropy - like categorical cross entropy but with integer labels instead of one-hot. |
LossType SquaredHinge |
Squared Hinge Loss - a smoother variant of hinge loss that squares the penalty. |
LossType Triplet |
Triplet Loss - learns embeddings using anchor, positive, and negative examples. |
LossType Wasserstein |
Wasserstein Loss - based on the Earth Mover’s Distance, commonly used in GANs. |
LossType WeightedCrossEntropy |
Weighted Cross Entropy Loss - cross entropy with class-specific weights. |
int value__ |